Implementation of CUDA-based loop mesh subdivision
Authors: Wang, M., Lv, H., Chang, J. and Zhang, J.J.
Conference: 2010 3rd International Conference on Computer and Electrical Engineering
Dates: 16-18 November 2010
Publisher: Institute of Electrical and Electronics Engineers, Inc.
Place of Publication: Chengdu, China
This work describes the implementation of Loop Subdivision on the GPU using CUDA. In order to achieve this, we have constructed face neighbours for Loop Subdivision using local triangulatation information, fully utilizing the single instruction, multiple threads (SIMT) and Pinned memory of GPU. This scheme can avoid the restrictions of traditional GPU programming. Traditional GPU applications are difficult to implement, maintain and are confined to the complexity of models. Our experiments show that real-time subdivision can be implemented for most of the models including the complex ones. Our proposed scheme could significantly speed up the computations.
Preferred by: Jian Chang and Jian Jun Zhang